GNU bug report logs - #16811
24.3.50; [ruby-mode] Implicit hash indentation bug

Previous Next

Full log

Message #16 received at 16811-done <at> debbugs.gnu.org (full text, mbox):

From: Dmitry Gutov <dgutov <at> yandex.ru>
To: Stefan Monnier <monnier <at> iro.umontreal.ca>
Cc: 16811-done <at> debbugs.gnu.org, Bozhidar Batsov <bozhidar <at> batsov.com>
Subject: Re: bug#16811: 24.3.50; [ruby-mode] Implicit hash indentation bug
Date: Sun, 23 Feb 2014 07:55:37 +0200

On 20.02.2014 16:11, Stefan Monnier wrote:
>> The problem seems to be that (let (smie--parent) (smie-indent--parent))
>
> Which does little more than (smie-backward-sexp 'halfsexp) in this case.

Yes.

>> called between "=>" and "{" doesn't stop at "=>", but goes straight to
>> the parent "{".
>
> Which seems correct.

Maybe according to the current grammar. But not if we want expressions 
after `=>' to have consistent additional indentation. Which I'm now 
thinking we probably don't, since otherwise, if it works right,

     {
       'HashSyntax' => {
         'EnforcedStyle'   => 'ruby19',
         'SupportedStyles' => %w(ruby19 hash_rockets)
       },

would turn into

     {
       'HashSyntax' => {
           'EnforcedStyle'   => 'ruby19',
           'SupportedStyles' => %w(ruby19 hash_rockets)
         },

and that doesn't look good. So indenting the left and right sides of 
`=>' to the same level is probably the way to go.

Changed in revision 116534, fixing this bug.

Thanks for the questions!

> This said, I don't understand why
>
>     {
>       'HashSyntax' => {
>         'EnforcedStyle'   => 'ruby19',
>         'SupportedStyles' => %w(ruby19 hash_rockets)
>       },
>       'SpaceAroundOperators' => { 'Enabled' => true }
>     }
>
> is indented differently from
>
>     {'HashSyntax' => {
>                       'EnforcedStyle'   => 'ruby19',
>                       'SupportedStyles' => %w(ruby19 hash_rockets)
>                     },
>      'SpaceAroundOperators' => {
>        'Enabled' => true }}
>
> It seems like "it jumps back to { instead of =>" is not the full explanation.

That's because we always insert an implicit semicolon after { at eol, 
and that's because it's hard to distinguish between a curly that's 
opening a hash and a curly than opens a curly block.

And we need those after block-opening curlies, otherwise some token on 
the first line of the block might consider the curly as its parent, 
align to it, and block-opening curlies themselves align to the beginning 
of the statement.

So

it("is too!") {
  bar
    .qux
}

turns into

it("is too!") {
  bar
  .qux
}

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.